We disrupt the market with our artificially intelligent algorithms.

ETH Zürich as a research site

The Swiss Federal Institute of Technology (ETH) Zurich is one of the world's leading research institutions in the field of artificial intelligence. This is one of the reasons why renowned technology companies such as IBM, Google and Microsoft develop their intelligent innovations and technologies in Switzerland.

Also our basic technology is the result from several years of research at ETH Zurich. As an official ETH spin-off company, we closely collaborate with the Chair of Management Information Systems to continuously improve our artificial intelligence.

In addition, due to our groundbreaking technology, we are supported by Innosuisse and the Swiss National Science Foundation in course of the BRIDGE program. The goal of this program is to anchor innovative technologies from basic research in business practice.

Supported by:

Three decades of experience in ERP process automation

Due to a strategic development partnership with an ERP software provider, we have 26 years of experience in the automation of ERP processes. With the help of our extensive industry-specific knowledge and the associated understanding of ERP processes, we are able to develop tailor-made solutions for manufacturing companies. Moreover, we were able to train our models from the very beginning with a large and high-quality database. Due to our generic algorithms, this knowledge can be easily transferred to a wide range of customer-specific document types.

Awarded Innovation Pioneer

The enormous potential of our technology has been awarded in various competitions as well, where we were always among the last 10% of participants. We are proud in particular for our success in the prestigious >>venture>> competition and our inclusion in the world-renowned accelerator program "MassChallenge Switzerland“. It shows that we are on the right way to revolutionize data extraction with our artificial intelligence. The grant from the Swiss Climate Foundation further demonstrates that our technology can also be used to have a lasting positive impact on the climate.

Unique system engineering approach

Our technology is based on a unique system engineering approach. By combining deep learning algorithms of Computer Vision and Natural Language Processing, we set new standards in machine learning. Especially, the reliable recognition of table positions represents a novelty in automated information extraction from documents. In contrast to traditional template and rule-based OCR systems, our Artificial Intelligence abstracts its knowledge from already processed documents and transfers it to new documents with different formats. The software also has an understanding of uncertainties for each individual extraction value. If such uncertainty occurs, the system collaborates with the human operator to correct the errors. Based on the corrections, the system immediately learns by its own. In order to facilitate the onboarding process and to guarantee the highest level of customer satisfaction, we have developed an auto-annotation algorithm. This algorithm automatically annotates documents based on existing database entries. Therefore, we are able to automatically onboard new customers with their data from the past while keeping the onboarding effort to a minimum. To protect our innovation, we are currently in the process of filing a patent application.

Human - Machine – Cooperation

Our artificially intelligent solution does not aim to completely execute entire processes on its own. From a technological point of view, it would not be a serious promise to achieve 100% extraction accuracy using machine learning. Especially for liquidity-relevant information, the human operator should always be in the loop having the overall control over the system’s decisions.

Moreover, our solution is not intended to replace employees, but to support them in their daily work. In fact, we want to enable them to shift their focus from mind-numbing, repetitive data entry tasks to more value-adding and meaningful tasks.

No black box AI

We work with an incremental approach of machine learning engineering. Using relatively small, annotated data sets, we create baseline algorithms which we then improve step by step. In doing so, we prevent that intelligent, but non-transparent black boxes are created. In contrast to other data extraction solutions with very deep neural networks, our solution always keeps the human in the loop who is able to understand decisions of the system and correct errors accordingly. This makes it possible to continuously improve the performance of our solution. The incremental approach also has the advantage that the microservices underlying the generalistic overall system can be easily extended to other problems and document types.